Mental replay rescales the subjective duration of past events via alpha and theta oscillations

Abstract

Human memory recall is flexible in time, allowing past events to be mentally replayed at variable speeds. However, it remains unclear whether temporally rescaled replay distorts the perceived duration of remembered events, and whether such distortion is tied to changes in the retrieved content. Here we instructed participants to mentally replay previously encoded naturalistic videos at compressed, expanded, or self-paced (free) speeds while recording scalp electroencephalography (EEG). Replaying an event at an expanded pace lengthened participants' subsequent judgments of its original duration relative to free and compressed replay, and the longer an event had been replayed, the longer it was judged to have lasted. Crucially, recalled memory content was unaffected, with gist accuracy and recalled detail comparable across speed conditions. During replay, alpha and theta power were reduced under compression, and both interacted with the speed instruction in predicting reported duration. Specifically, alpha power exhibited a stronger positive relationship with duration judgments under compressed replay than under free or expanded replay. Conversely, the relationship between theta power and reported duration was significantly more positive under expanded replay than under free or compressed replay. Neural pattern similarity between instructed and later spontaneous replay varied with replay speed, yet this reinstatement neither predicted reported duration nor mediated the oscillatory effects. These findings suggest that remembered duration is not read out from the amount of content retrieved but reconstructed at retrieval through oscillatory dynamics, decoupling the subjective scaling of time from the content of memory.